The research target is to explore the key biomarkers of colorectal cancer by studying the impact of dietary factors on colorectal cancer. We first employed statistical methods to preprocess experimental data. Then, relief algorithm is employed to extract key features in the dietary data set. Finally, supporting vector machine (SVM) is used to classify the data set and compute classification accuracy. The results demonstrated that vegetables, seafood, eggs and milk have great impacts on the colorectal cancer. Therefore, we concluded that integrating relief algorithm with SVM model can explore the key biomarkers for colorectal cancer, while the investigation of the interactions among these features needs further research.